Skip to content

/encyclopedia-writing

Encyclopedic neutral-reference writing craft — NPOV (attribute facts not opinions, due weight, neutral faction labels, verdict restraint), summary style and Coatrack avoidance, wikilink conventions (link density, first-mention, slug alias, abbreviation glossing). Use when

From plugin
706 skills6 agents10 commands
shell
$ npx -y skills add alfadur7/llm-wiki-newsroom --skill encyclopedia-writing --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/encyclopedia-writing
How auto-invocation works

Context preview

The summary Claude sees to decide when to auto-load this skill.

Encyclopedic neutral-reference writing craft — NPOV (attribute facts not opinions, due weight, neutral faction labels, verdict restraint), summary style and Coatrack avoidance, wikilink conventions (link density, first-mention, slug alias, abbreviation glossing). Use when

SKILL.md

encyclopedia-writing.SKILL.md
name: encyclopedia-writing
description: Encyclopedic neutral-reference writing craft — NPOV (attribute facts not opinions, due weight, neutral faction labels, verdict restraint), summary style and Coatrack avoidance, wikilink conventions (link density, first-mention, slug alias, abbreviation glossing). Use when writing or reviewing a neutral encyclopedic/wiki reference, landscape overview, hub, or source page, or when a source-based connected document that advocates no particular viewpoint is needed.

encyclopedia-writing

Neutral-reference writing craft drawn from the encyclopedic editing tradition (the Wikipedia policy family). It keeps this wiki's overview, issue, and hub pages a connected reference that is grounded in facts and sources and advocates no particular viewpoint. `criteria.json` is the SoT for each criterion's definition, comparator, and source; content-type thresholds and algorithm params are injected by `.claude/layers/_manifest.json` (the skill is content-type-agnostic). Examples are illustrative of the target English prose; the Korean-rendering rules below apply only when WIKI_LANG=ko and are inert on the default English corpus.

Wikilink connectivity (enc.link-density · enc.first-mention · enc.lead-body)

A reference's value comes from the link density that lets a reader move to adjacent concepts; at the same time, re-linking the same target within one section is visual noise.

  • **enc.link-density** — the EDITOR region (excluding AUTO blocks) holds at least the threshold number of `[[wikilink]]`s. "Articles on highly technical subjects might demand a higher density of links," so technical/finance domains are allowed slightly above the ceiling.
  • **enc.first-mention** — "link a term at most once per major section, at first occurrence." A second link of the same stem within a section becomes plain text. A section that structurally cites the same target twice (e.g. adjacent-field boundary prose) is exempted via the manifest's `exempt_sections`.
  • **enc.lead-body** — "the lead of an article usually has a greater density of links than later parts of the article"; this is expected, since the reader's first encounter is where navigation options should be densest. Because which span counts as the lead is content-type-bound, the orchestrator measures it.

All three are deterministic; `checks.py` (link-density·first-mention) or the orchestrator (lead-body) computes PASS/FAIL with manifest-injected thresholds/params.

Neutrality & notation (enc.verdict-restraint · enc.slug-alias · enc.abbr-gloss · enc.encyclopedic-tone)

A reference advocates no viewpoint (NPOV) and follows readable notation conventions (MoS). Deterministic; thresholds injected by manifest.

  • **enc.verdict-restraint** — "avoid stating opinions as facts" and "describe disputes, but do not engage in them." Write evaluative sentences in three strengths — observation ("the metric points the opposite way") and conditional recommendation ("appears to be"·"may be") are fine, but a verdict ("is reasonable"·"has the most explanatory power"·"is consistent with") over the threshold must be softened with hedging/attribution. Repeated verdicts under hedged wording stay neutral in form but biased in substance. (ko-localization renderings: observation `~ 지표가 정반대를 가리킨다`; conditional `~로 보인다`·`~일 가능성이 있다`; verdict `~이 합리적이다`·`가장 설명력이 높다`·`~에 부합한다`.)
  • **enc.slug-alias** — do not expose a long raw kebab-case identifier; show a human-readable alias (MoS overlinking/readability). This wiki uses English aliases by default; native-script aliases only under WIKI_LANG=ko.
  • **enc.abbr-gloss** — "an acronym should be written out in full for the first time, followed by the abbreviation in parentheses" (commonly used abbreviations such as global brands are exempt). Applies to abbreviations appearing in body text.
  • **enc.encyclopedic-tone** — keep the encyclopedic register; do not self-reference the editorial artifact or editorial decisions (WP:SELFREF — an article "shouldn't refer to [the work] in a non-neutral fashion"). Editorializing such as "this document treats … as"·"here we keep only …" is deleted or absorbed into a factual statement. (ko-localization renderings: `본 문서는 ~로 둔다`·`여기서는 ~만 유지한다`.)

Issue neutrality & balance (enc.npov-asf · enc.due-weight · enc.label-neutral · enc.back-reference)

A piece covering a dispute advocates no side and fairly juxtaposes both, grounded in facts and sources (WP:NPOV — ASF and DUE run through every section).

  • **enc.npov-asf** — attribute every value-judgment/interpretation/claim with a named subject + source + figure/quote (≥2 of the 3); "avoid stating opinions as facts," and use no weasel words ("some say"·"many people"·"as is known"; ko renderings `혹자는`·`많은 사람이`·`알려진 바로는`). judge=M. e.g. ✅ "The IPCC warned of coastal-city risk in its 2021 report" (named source + date) / ❌ "Many experts believe coastal cities are at risk" (anonymous plurality — weasel)
  • **enc.due-weight** — coverage is proportionate to prominence in reliable sources, not skewed to one side (WP:DUE); a minority view does not get equal space and is explicitly framed as a minority. judge=M. e.g. ✅ "The Earth is spherical and the scientific consensus is overwhelming; a minority dissents but without credible grounds" (proportionate weight + minority noted) / ❌ "The Earth is round. But some believe it is flat." (false balance)
  • **enc.label-neutral** — both faction labels (or the aggregate's tension-axis titles) must be drawn from an equivalent vocabulary set. If only one side gets a value-laden word ("myth"·"empirical proof"·"skepticism"·"innovation," etc.), that side is privileged or demoted — unify both on a structural/standpoint basis ("replacement vs augmentation"·"industry-led vs regulation-led"). Deterministic check for one-sided skew = 0. **ko-localization rendering rule (WIKI_LANG=ko only)** — even when the English original is neutral, a literal Korean rendering can introduce connotation, so when loc
Read more
Read it on GitHub ↗

Showing the first part of this file.

Ships withllm-wiki-newsroom

A multi-agent AI knowledge base run by a five-role "newsroom" — open-source, local-first, no vendor lock-in.

Get the whole plugin, auto-invoked
Stats
70
Stars
0
Views
11
Forks
Active
Maintenance
Python
Language
MIT
License
1d ago
Last commit
1mo ago
Created

Repo: alfadur7/llm-wiki-newsroom

Other skills on llm-wiki-newsroom.